Close this search box.

Gen 6 (32Gb) FC is All About Enabling More Efficient and Effective SAN Administration; Interview with QLogic’s Vikram Karvat, Part 3

Mapping worldwide names (WWNs) to LUNs and doing recurring rezoning in Fibre Channel (FC) SANs is a reality that every SAN administrator deals with on a regular basis. However the latest features found in Gen 6 (32Gb) FC offers new hope for these individuals by making these jobs simpler and easier to perform. In this third and final installment in my interview series with QLogic’s Vice President of Products, Marketing and Planning, Vikram Karvat, he provides some insight into the multiple new features that Gen 6 FC offers to help SAN and storage administrators perform their jobs more efficiently and effectively.
gen-6 FC
Jerome: What does Gen 6 FC do to help SAN and storage administrators perform their jobs more effectively?
Vikram: Gen 6 helps these individuals perform their jobs more effectively in three ways.
First is it does improve performance especially for mission critical applications, both from a bandwidth as well as an IOPS perspective.
Second, we already know how to build highly resilient infrastructures and we make them even more resilient with some of the features that we implement around Forward Error Correction (FEC) and quality of service (QoS).
Third, some of the features and functions focus on reducing OPEX. When I refer to OPEX, think about it cohesively, all the way from deploying a new FC SAN, being able to manage it over its life cycle, and, if you have issues, being able to diagnose and resolve those issues quickly.
Our data centers today are not static. They require administrators, over the course of time, to more effectively migrate data from one physical host to another. That is where a lot of our effort over the last year has been to reduce the overall life cycle costs associated with deploying and managing FC in addition to just getting Gen 6 (32Gb) FC functionality out the door.
Jerome: Please tell me about your new FC HBAs and what were some of the drivers behind their new features?
Vikram: We are announcing s a family of adapters as well as controllers to embed on the storage array side in single, dual and quad port configurations. We have invested in developing Gen 5 (16Gb) FC and Gen 6 (32Gb) FC monolithic (i.e. single chip) quad port solutions, a first for QLogic and for the industry in general. Our decision to develop these monolithic quad-port solutions was driven by two factors

  1. Demand on the storage array side for increased port density.
  2. We are, interestingly enough, actually seeing increased quad-port demand on the server side.

We are especially seeing this demand in the service provider market. For example, some of the large service providers are deploying virtualized, multi-tenant type of infrastructures. They are looking for four and sometimes eight FC ports coming off a single server, primarily to enable performance, as well as essentially segregate some of their storage networks via physical ports in multi-tenant environments. That’s definitely new. That use case actually shows where I think FC is going. It is moving from a traditional enterprise into more off premise, virtualized, multi-tenant data center types of deployments.
Related to this, back in October 2015 we announced the StorFusion initiative that is focused around orchestration and management services. From a manageability perspective and TCO perspective, one of the biggest issues we run into is faults with either the cabling plant or optics. Even after all these years it’s the simplest things that tend to screw things up. So as we get into denser environments with ever more complex patch panels, this became more and more of a problem.
To address this, we developed a feature with Brocade called Clear Link diagnostic port (D port) that effectively allows you to look at the cable plant and optics from an end to end perspective and isolate the source if a fault has occurred. It can also help identify a degradation in performance which may be indicative of a fault that may occur. With this feature, we can localize it to a particular optical module, switch port or a particular length of cable. We effectively provide the IT user a mechanism to do preventative maintenance in these dense environments before issues arise.
In addition to identifying issues, we said, “OK, wouldn’t it be cool if you were able to provide some light path diagnostics?” In other words, if service personnel actually showed up, they would actually know what to unplug, and avoid a mistake by unplugging the wrong cable from the port.
To address this concern, we implemented a feature called link cable beaconing (LCB). This frees an IT manager from a central console to toggle the light on the target port on and off. The port will merrily blink until you do something about it. Is this the sexiest feature on the face of the planet? No. But it provides real value to the people that use FC every day.
Another feature that we implemented, again with Brocade, is a feature called fabric-based boot LUN discovery (F-BLD). Anytime you bring up a new host, you have umpteen steps you are required to follow before you can get it onto the SAN and boot it up. Further, in some data centers, administrators simultaneously bring up in an entire rack of hosts and wire them up. That takes a significant amount of time and effort.
F-BLD actually pre-provisions the boot LUN information within the fabric. When I set up my fabric, or if I know that I am going to roll in 100 hosts, I can go in and pre-provision all of the LUNs. Then when the server boots up, it’ll automatically configure itself and begin booting by going into the fabric and picking up its boot factors.
This feature can reduce the number of steps that it takes to get a FC host live on a SAN by up to 30 percent. It does not eliminate all of the steps, but 30 percent fewer steps is 30 percent fewer steps. Again, this goes back to the principle of, “How do we help IT become more efficient?” There are certain tasks they have to do and cannot get around. We cannot fix all of the problems but we can make certain tasks more efficient.
Finally, another feature that we have implemented is fabric-assigned worldwide names (FA-WWN). Right now there is a worldwide name (WWN) that is associated with a given port. The WWN is wired into the port and is the identity of the port forever. FA-WWN provides a virtual address for that port. The advantage is that it decouples zoning, LUN configurations, access control lists (ACLs), etc. that are associated with that WWN from the actual physical server.
Now if you have to move the image that is running on one physical host to another physical host, you can move that virtual WWN with it. In a SAN environment you get much simpler migration within your data centers as you do not have to keep going into your FC switch to redo zoning every time. Again you only care about this if you are the one having to rezone your SAN infrastructure.
In part 1 of this interview series, Vikram Karvat, provides some background as to why FC remains relevant and how all-flash arrays are one of the forces driving the need for 32Gb FC.
In part 2 of this interview series, Vikram Karvat, explains why 32Gb FC HBAs are still installed separately in servers and provides insight into what new features may be released in the Gen 7 FC protocol.


Click Here to Signup for the DCIG Newsletter!


DCIG Newsletter Signup

Thank you for your interest in DCIG research and analysis.

Please sign up for the free DCIG Newsletter to have new analysis delivered to your inbox each week.